49.4k- Resources
- Engineering
ClickHouse engineering resources
What is eBPF? Kernel-level observability without code changes
eBPF runs sandboxed programs inside the Linux kernel to capture traces, network flows, and profiles without changing application code. How it works, where it falls short, and how to query its output at scale.
Last updated: Jul 29, 2026
What is distributed tracing? Traces, spans, and context propagation
Distributed tracing follows one request across every service it touches, recording each step as a span under a shared trace ID so you can see where the time went.
Last updated: Jul 29, 2026
What is AIOps, and what does it actually deliver?
AIOps applies machine learning to IT operations data to detect anomalies, correlate events, and cut alert noise. What it does well, where it stops, and what it needs underneath.
Last updated: Jul 29, 2026
What is AI agent observability?
AI agent observability traces multi-step, non-deterministic agent workflows (LLM calls, tool calls, retrievals, sub-agent spawns) as spans in one trace, so teams can debug agent behavior.
Last updated: Jul 29, 2026
The three pillars of observability - and where the model breaks
Logs, metrics, and traces defined honestly: what each pillar captures, when to use which, how they correlate, and why siloed pillars break down at scale.
Last updated: Jul 29, 2026
What is tail latency? Why p99 spikes when averages look fine
Tail latency is the slow end of the response-time distribution (p95 and beyond). Why fan-out makes rare slowness common, what causes it, and how hedged requests, timeouts, and load shedding reduce it.
Last updated: Jul 29, 2026
SLOs, SLIs, and SLAs: definitions, differences, and worked examples
An SLI measures a service, an SLO sets the internal target for that measurement, and an SLA is the external contract with penalties. How the three layers fit together, with error-budget math and worked examples.
Last updated: Jul 29, 2026
RUM vs synthetic monitoring: which do you need?
RUM measures what real users experienced; synthetic monitoring measures what scripted probes experienced. How the two differ on alerting, coverage, and cost, and when to run both.
Last updated: Jul 29, 2026
What is root cause analysis? Methods, process, and limits
Root cause analysis traces an incident from symptom to underlying cause. The methods, a worked example, a report template, and where the single-root-cause model breaks down.
Last updated: Jul 29, 2026
RED method vs USE method: which should you use?
RED (Rate, Errors, Duration) monitors request-driven services; USE (Utilization, Saturation, Errors) monitors resources. How they differ, a worked outage through both lenses, and why most teams run both.
Last updated: Jul 29, 2026
Percentiles vs averages: why your latency dashboard lies
Latency distributions are right-skewed, so the mean hides the slow tail users actually feel. What p50, p95, and p99 mean, why you cannot average percentiles, and how to compute them at scale.
Last updated: Jul 29, 2026
OpenTelemetry's profiles signal enters public alpha
OpenTelemetry profiles, the fourth telemetry signal, entered public alpha in March 2026, with OTLP support, collector pipelines from v0.148.0, and a whole-system eBPF profiler.
Last updated: Jul 29, 2026
OpenTelemetry semantic conventions explained
OpenTelemetry semantic conventions standardize telemetry attribute names like http.request.method and db.system.name, so data from any service is portable and queryable across tools.
Last updated: Jul 29, 2026
OpenTelemetry histograms explained: buckets, exponential histograms, and percentiles
How OpenTelemetry histograms summarize distributions: explicit-bucket vs base-2 exponential encodings, how percentiles are computed from buckets and their error bounds, and how histogram data is stored and queried.
Last updated: Jul 29, 2026
Observability vs visibility: the terms teams conflate
Visibility means seeing predefined surfaces; observability means interrogating internal state for unknowns. Why network teams say one and software teams say the other.
Last updated: Jul 29, 2026